- in the downloading code, handle 404 errors more gently. Place information that given file was not available due to broken link.
- add more unit tests for different scenarios of algorithm and tree state e.g.:
- tree with the root only
- a large tree
- tree with no missing nodes and without errors
- root's children are leafs
- in the Unit tests, when the mechanize.Browser opens the "/root/2011-07-02" webpage, the "HTTP Error 404: File not found" error sometimes shows up. The entry in the logs looks as follows:
WARNING thread=="Thread-95", exception in node "/root/2011-07-12": "Moving to child failed". Detailed message: "HTTP Error 404: File not found"..- probably a fault of the HTTP file server?
- ? change the order of nodes saved in the
*.xmlfile: order by names, not by state of the node? - add command-line option to try to visit once again the ERROR nodes of a tree read from a file (it might be helpful because in case of a web crawler, the pages which were not available before might be available now)
? generate source docs with sphinx instead of epydoc?
- In the
docs/algorithm/index.texdocument,- make a distinction between tree as a data structure shared between the threads and explored tree
- describe more precisely the example of the upward update of the state (the last figure).